The lookup plugin is always executed on the control node (that's your Ansible server), which is to say that the lookup plugin does not do something on your managed nodes (the target systems).

To use the pipe plugin, the Python pipe.py script must be in your Ansible plugins lookup directory, such as /usr/lib/python2.7/site-packages/ansible/plugins/lookup/pipe.py.

The lookup plugin can be used to return the hostname of your control node (your Ansible server). Let's say for example you have the following playbook.

---
- hosts: all
  tasks:
    - name: return control node hostname
      debug:
        msg: "{{ lookup('pipe', 'hostname') }}"
      run_once: true
...

Running this playbook should return something like this. In this example, ansible.example.com is the hostname of the control node.

TASK [return control node hostname]
ok: [server1.example.com] => {
    "msg: "ansible.example.com"
}

Or, there are a number of different magic variable will return the hostname of a managed node (e.g. the target systems).

• ansible_hostname will return the hostname of a managed node from facts
• ansible_play_batch will return the hostname of a managed node, just like the Linux hostname command, in the form of an array of strings.
• ansible_play_hosts_all will return the hostname of a managed node, just like the Linux hostname command, in the form of an array of strings.
• ansible_limit will return the hostname passed to the -l or --limit command line option.
• inventory_hostname will return the hostname of a manage node as listed in your default hosts file or your own inventory file.
• inventory_hostname_short will return the short hostname of a manage node as listed in your default hosts file or your own inventory file.
• play_hosts will return the hostname of a managed node, just like the Linux hostname command, in the form of an array of strings.